Evidence-Driven Reasoning for Industrial Maintenance Using Heterogeneous Data

Fearghal O'Donncha; Nianjun Zhou; Natalia Martinez; James T Rayfield; Fenno F. Heath; Abigail Langbridge; Roman Vaculin

cs.AIMar 9, 2026

Evidence-Driven Reasoning for Industrial Maintenance Using Heterogeneous Data

Authors: Fearghal O'Donncha, Nianjun Zhou, Natalia Martinez, James T Rayfield, Fenno F. Heath, Abigail Langbridge, Roman Vaculin

Organizations: IBM Research Europe, Ireland · IBM Research Yorktown, USA · Imperial College London, UK

Abstract

Industrial maintenance platforms contain rich but fragmented evidence, including free-text work orders, heterogeneous operational sensors or indicators, and structured failure knowledge. These sources are often analyzed in isolation, producing alerts or forecasts that do not support conditional decision-making: given this asset history and behavior, what is happening and what action is warranted? We present Condition Insight Agent, a deployed decision-support framework that integrates maintenance language, behavioral abstractions of operational data, and engineering failure semantics to produce evidence-grounded explanations and advisory actions. The system constrains reasoning through deterministic evidence construction and structured failure knowledge, and applies a rule-based verification loop to suppress unsupported conclusions. Case studies from production CMMS deployments show that this verification-first design operates reliably under heterogeneous and incomplete data while preserving human oversight. Our results demonstrate how constrained LLM-based reasoning can function as a governed decision-support layer for industrial maintenance.

Figures & tables

Figure 1: End - to - end pipeline of the Condition Insight System. Heterogeneous asset data and failure - mode knowledge are consolidated into structured evidence packets, which are processed by a domain LLM agent with verifier - based checking to produce a UI - agnostic condition insight summary.

Figure 2: Example condition insight generated by the proposed framework using historical work orders of the last 200 days for a motor asset. The output separates evidence-based observations (top) from prioritized, actionable recommendations (bottom), synthesizing maintenance history, asset metadata, health indicators, and data consistency checks into a decision-oriented summary.

Prompt	Scope	UCR $\downarrow$	HSR $\uparrow$	CAR $\uparrow$	MIC
Naive	WO	0.007	0.64	0.68	4.9
Naive	All	0.003	0.66	0.70	4.9
Constr.	WO	0.002	0.68	0.89	3.2
Constr.	All	0.008	0.71	0.91	3.3

Table 1: Effect of evidence enrichment and prompt constraints using GPT-OSS evaluated by Claude-4. Prompt constraints substantially improve rule compliance (CAR) and reduce output verbosity (MIC), while structured evidence improves grounding under naive prompting.

Explore similar work

IndustryAssetEQA: A Neurosymbolic Operational Intelligence System for Embodied Question Answering in Industrial Asset Maintenance
Apr 25, 2026Chathurangi Shyalika, Dhaval Patel, Amit ShethEmbodied QA Predictive Maintenance
Artificial Intelligence Institute, University of South Carolina USA · Software Innovation Lab IBM Yorktown USA · Artificial Intelligence Institute, University of South Carolina Indian AI Research Organization
DiagnosticIQ: A Benchmark for LLM-Based Industrial Maintenance Action Recommendation from Symbolic Rules
May 9, 2026Devin Yasith De Silva, Dhaval Patel, Christodoulos Constantinides +7Logical Reasoning LLM Evaluation
2Rensselaer Polytechnic Institute · 1IBM · 3Independent Contributor
FailureSensorIQ: A Multi-Choice QA Dataset for Understanding Sensor Relationships and Failure Modes
Jun 3, 2025Christodoulos Constantinides, Dhaval Patel, Shuxin Lin +3Question Answering LLM Reliability
IBM · IBM TJ Watson Research Center

Apr 25, 2026J/K move · Enter open · S save

IndustryAssetEQA: A Neurosymbolic Operational Intelligence System for Embodied Question Answering in Industrial Asset Maintenance

Chathurangi Shyalika, Dhaval Patel, Amit Sheth

Artificial Intelligence Institute, University of South Carolina USA · Software Innovation Lab IBM Yorktown USA · Artificial Intelligence Institute, University of South Carolina Indian AI Research Organization

Embodied QA Predictive Maintenance Knowledge Graph Question Answering Neuro-Symbolic AI

Open paper AskPDF arXiv

Industrial maintenance environments increasingly rely on AI systems to assist operators in understanding asset behavior, diagnosing failures, and evaluating interventions. Although large language models (LLMs) enable fluent natural-language interaction, deployed maintenance assistants routinely produce generic explanations that are weakly grounded in telemetry, omit verifiable provenance, and offer no testable support for counterfactual or action-oriented reasoning that undermine trust in safety-critical settings. We present IndustryAssetEQA, a neurosymbolic operational intelligence system that combines episodic telemetry representations with a Failure Mode Effects Analysis Knowledge Graph (FMEA-KG) to enable Embodied Question Answering (EQA) over industrial assets. We evaluate on four datasets covering four industrial asset types, including rotating machinery, turbofan engines, hydraulic systems, and cyber-physical production systems. Compared to LLM-only baselines, IndustryAssetEQA improves structural validity by up to 0.51, counterfactual accuracy by up to 0.47, and explanation entailment by 0.64, while reducing severe expert-rated overclaims from 28% to 2% (approximately 93% reduction). Code, datasets, and the FMEA-KG are available at https://github.com/IBM/AssetOpsBench/tree/IndustryAssetEQA/IndustryAssetEQA.

Abstract

Explore similar work

IndustryAssetEQA: A Neurosymbolic Operational Intelligence System for Embodied Question Answering in Industrial Asset Maintenance

DiagnosticIQ: A Benchmark for LLM-Based Industrial Maintenance Action Recommendation from Symbolic Rules

FailureSensorIQ: A Multi-Choice QA Dataset for Understanding Sensor Relationships and Failure Modes